def test_as_training_data_error(self):
with self.assertRaises(ValueError):
feature = IndexedPairFeature([
IndexedFeatureWord(1, [1, 2]),
IndexedFeatureWord(4, [1, 2, 6])
], [IndexedFeatureWord(1, [1, 2]),
IndexedFeatureWord(3, [5])], None)
feature.as_training_data()
with self.assertRaises(ValueError):
self.feature.as_training_data(mode="words+character")
class TestIndexedPairFeature(DuplicateTestCase):
def setUp(self):
super(TestIndexedPairFeature, self).setUp()
self.feature = IndexedPairFeature([
IndexedFeatureWord(1, [1, 2]),
IndexedFeatureWord(2, [3, 4]),
IndexedFeatureWord(3, [5]),
IndexedFeatureWord(5, [1, 4, 1]),
IndexedFeatureWord(4, [1, 2, 6])
], [
IndexedFeatureWord(1, [1, 2]),
IndexedFeatureWord(8, [3, 1, 2, 1]),
IndexedFeatureWord(2, [3, 4]),
IndexedFeatureWord(3, [5])
], [0, 1])
def test_get_lengths(self):
assert self.feature.get_lengths() == {
"num_sentence_words": 5,
'num_word_characters': 4
}
def test_pad_adds_padding_words(self):
self.feature.pad({"num_sentence_words": 6, 'num_word_characters': 5})
first_sent_word_idxs, second_sent_word_idxs = self.feature.get_int_word_indices(
)
first_sent_char_idxs, second_sent_char_idxs = self.feature.get_int_char_indices(
)
assert first_sent_word_idxs == [1, 2, 3, 5, 4, 0]
assert second_sent_word_idxs == [1, 8, 2, 3, 0, 0]
assert first_sent_char_idxs == [[1, 2, 0, 0, 0], [3, 4, 0, 0, 0],
[5, 0, 0, 0, 0], [1, 4, 1, 0, 0],
[1, 2, 6, 0, 0], [0, 0, 0, 0, 0]]
assert second_sent_char_idxs == [[1, 2, 0, 0, 0], [3, 1, 2, 1, 0],
[3, 4, 0, 0, 0], [5, 0, 0, 0, 0],
[0, 0, 0, 0, 0], [0, 0, 0, 0, 0]]
assert self.feature.label == [0, 1]
def test_pad_truncates(self):
self.feature.pad({"num_sentence_words": 2, 'num_word_characters': 3})
first_sent_word_idxs, second_sent_word_idxs = self.feature.get_int_word_indices(
)
first_sent_char_idxs, second_sent_char_idxs = self.feature.get_int_char_indices(
)
assert first_sent_word_idxs == [1, 2]
assert second_sent_word_idxs == [1, 8]
assert first_sent_char_idxs == [[1, 2, 0], [3, 4, 0]]
assert second_sent_char_idxs == [[1, 2, 0], [3, 1, 2]]
assert self.feature.label == [0, 1]
def test_pad_general(self):
self.feature.pad(self.feature.get_lengths())
first_sent_word_idxs, second_sent_word_idxs = self.feature.get_int_word_indices(
)
first_sent_char_idxs, second_sent_char_idxs = self.feature.get_int_char_indices(
)
assert first_sent_word_idxs == [1, 2, 3, 5, 4]
assert second_sent_word_idxs == [1, 8, 2, 3, 0]
assert first_sent_char_idxs == [[1, 2, 0, 0], [3, 4, 0,
0], [5, 0, 0, 0],
[1, 4, 1, 0], [1, 2, 6, 0]]
assert second_sent_char_idxs == [[1, 2, 0, 0], [3, 1, 2, 1],
[3, 4, 0, 0], [5, 0, 0, 0],
[0, 0, 0, 0]]
assert self.feature.label == [0, 1]
def test_as_training_data_produces_correct_numpy_arrays(self):
self.feature.pad({'num_sentence_words': 3, 'num_word_characters': 2})
inputs, label = self.feature.as_training_data()
assert_allclose(label[0], np.asarray([0, 1]))
assert len(inputs) == 2
assert_allclose(inputs[0], np.asarray([1, 2, 3]))
assert_allclose(inputs[1], np.asarray([1, 8, 2]))
inputs, label = self.feature.as_training_data(mode="character")
assert_allclose(label[0], np.asarray([0, 1]))
assert len(inputs) == 2
assert_allclose(inputs[0], np.asarray([[1, 2], [3, 4], [5, 0]]))
assert_allclose(inputs[1], np.asarray([[1, 2], [3, 1], [3, 4]]))
inputs, label = self.feature.as_training_data(mode="word+character")
assert_allclose(label[0], np.asarray([0, 1]))
assert len(inputs) == 4
assert_allclose(inputs[0], np.asarray([1, 2, 3]))
assert_allclose(inputs[1], np.asarray([[1, 2], [3, 4], [5, 0]]))
assert_allclose(inputs[2], np.asarray([1, 8, 2]))
assert_allclose(inputs[3], np.asarray([[1, 2], [3, 1], [3, 4]]))
def test_as_training_data_error(self):
with self.assertRaises(ValueError):
feature = IndexedPairFeature([
IndexedFeatureWord(1, [1, 2]),
IndexedFeatureWord(4, [1, 2, 6])
], [IndexedFeatureWord(1, [1, 2]),
IndexedFeatureWord(3, [5])], None)
feature.as_training_data()
with self.assertRaises(ValueError):
self.feature.as_training_data(mode="words+character")
def test_as_testing_data_produces_correct_numpy_arrays(self):
self.feature.pad({'num_sentence_words': 4, 'num_word_characters': 2})
inputs, labels = self.feature.as_testing_data()
assert len(labels) == 0
assert len(inputs) == 2
assert_allclose(inputs[0], np.asarray([1, 2, 3, 5]))
assert_allclose(inputs[1], np.asarray([1, 8, 2, 3]))
inputs, label = self.feature.as_training_data(mode="character")
assert len(labels) == 0
assert len(inputs) == 2
assert_allclose(inputs[0], np.asarray([[1, 2], [3, 4], [5, 0], [1,
4]]))
assert_allclose(inputs[1], np.asarray([[1, 2], [3, 1], [3, 4], [5,
0]]))
inputs, label = self.feature.as_training_data(mode="word+character")
assert len(labels) == 0
assert len(inputs) == 4
assert_allclose(inputs[0], np.asarray([1, 2, 3, 5]))
assert_allclose(inputs[1], np.asarray([[1, 2], [3, 4], [5, 0], [1,
4]]))
assert_allclose(inputs[2], np.asarray([1, 8, 2, 3]))
assert_allclose(inputs[3], np.asarray([[1, 2], [3, 1], [3, 4], [5,
0]]))
def test_as_testing_data_error(self):
with self.assertRaises(ValueError):
self.feature.as_testing_data(mode="words+character")
def test_equals(self):
feature_1 = IndexedPairFeature(
[IndexedFeatureWord(1, [1, 2]),
IndexedFeatureWord(4, [1, 2, 6])],
[IndexedFeatureWord(1, [1, 2]),
IndexedFeatureWord(3, [5])], None)
feature_2 = IndexedPairFeature(
[IndexedFeatureWord(1, [1, 2]),
IndexedFeatureWord(4, [1, 2, 6])],
[IndexedFeatureWord(1, [1, 2]),
IndexedFeatureWord(3, [5])], None)
feature_3 = IndexedPairFeature(
[IndexedFeatureWord(1, [1, 2]),
IndexedFeatureWord(1, [2, 2])],
[IndexedFeatureWord(1, [1, 2]),
IndexedFeatureWord(3, [5])], None)
feature_4 = IndexedPairFeature([IndexedFeatureWord(1, [1, 2])],
[IndexedFeatureWord(1, [2, 2])], None)
self.assertNotEquals(feature_1, feature_4)
self.assertNotEquals(feature_1, feature_3)
self.assertFalse(feature_1.__eq__(0))
self.assertEquals(feature_1, feature_2)
def test_less_than(self):
feature_1 = IndexedPairFeature(
[IndexedFeatureWord(1, [1, 2]),
IndexedFeatureWord(4, [1, 2, 6])],
[IndexedFeatureWord(1, [1, 2]),
IndexedFeatureWord(3, [5])], None)
feature_2 = IndexedPairFeature(
[IndexedFeatureWord(1, [1, 2]),
IndexedFeatureWord(4, [1, 2, 6])],
[IndexedFeatureWord(2, [2, 2]),
IndexedFeatureWord(3, [5])], None)
feature_3 = IndexedPairFeature([IndexedFeatureWord(1, [1, 2])],
[IndexedFeatureWord(1, [2, 2])], None)
self.assertFalse(feature_1.__lt__(0))
self.assertFalse(feature_2.__lt__(feature_1))
self.assertLess(feature_1, feature_2)
self.assertLess(feature_3, feature_2)